Design and Analysis of Sequential, Multiple Assignment, Randomized Trials

A workshop to introduce adaptive interventions and sequential, multiple assignment, randomized trials (SMARTs) guided by examples. We unpack the black box of multi-component, sequential interventions and present appropriate aims and analyses for SMART design. This workshop will help you write a grant to get a SMART funded and/or learn analytic methods to apply to SMART data.

Instructor:
Kelley Kidwell, PhD (University of Michigan)

Workshop Dates and Times:
Monday, June 16, 2025 at 10:00am-3:30pm EDT
Tuesday, June 17, 2025 at 10:00am-3:30pm EDT
(And 2024 Recording is Available Now)

Workshop Format:
Two-Day Online Workshop and Recording

If only a single pill could cure all ailments. In reality, most diseases require ongoing, personalized treatment that adapts over time. However, the strongest evidence for treatment effectiveness often comes from traditional randomized controlled trials (RCTs), which evaluate an intervention at a single point in time. While valuable, RCTs may overlook how treatments interact over time, missing key synergies or antagonisms. To address this, adaptive interventions—also called dynamic treatment regimens—guide whether, how, and when to adjust treatment at critical decision points. Clinicians often rely on judgment and piecemeal evidence from point-in-time RCTs, which can introduce bias. A more rigorous way to develop these regimens is through a sequential, multiple assignment, randomized trial (SMART).

A SMART is a multi-stage trial in which participants are randomized at least twice, with subsequent treatment assignments based on prior responses or other participant or disease characteristics. Each SMART embeds multiple dynamic treatment regimens, allowing researchers to estimate the effects of tailored treatment sequences. SMARTs have been applied across various fields, including cancer, mental health, HIV, chronic pain, tobacco cessation, education, and implementation science. While most use individual randomization, some employ cluster randomization, especially in implementation science or disease prevention. Numerous funding agencies, including the NIH and PCORI, have supported SMART designs, and methods exist for diverse trial structures and outcome types.

This workshop introduces dynamic treatment regimens and explains how SMART designs can develop high-quality, evidence-based interventions over several critical treatment decisions. We begin with an overview of dynamic treatment regimens—defining key components, discussing their relevance, and reviewing examples from different fields. Next, we introduce SMART design, illustrating its principles through real-world case studies that highlight different trial features. We compare SMARTs to similar designs and discuss their primary, secondary, and exploratory aims, with a focus on grant proposals and trial protocols. The workshop concludes with a high-level overview of SMART analysis, including power and sample size calculations, as well as an introduction to small sample SMARTs, which efficiently estimate first-stage treatment effects with limited participants. Methods will be demonstrated through case studies and hands-on coding in R.

What you’ll learn

  • Adaptive interventions: understand the definition and components of adaptive interventions

  • SMART design: Define SMART design principles, primary aims, and advantages; Compare & contrast SMART to other designs; Illustrate designs through examples of funded designs

  • Analytic Methods: Apply regression methods to estimate embedded adaptive interventions using R software.

  • Grant writing skills: tips and tricks when proposing a SMART design in grant proposal

Syllabus

Introduction to Adaptive Interventions

  1. What are adaptive interventions?

  2. What are the components of an adaptive intervention?

  3. Why are adaptive interventions needed?

  4. What are design goals of adaptive interventions?

Introduction to SMARTs

  1. What are SMARTs?

  2. SMARTs vs. other trial designs

  3. Alternative Approaches to a SMART & Case Studies

SMART Design Principles & Aims

  1. SMART design principles

  2. Typical primary, secondary, and exploratory aims in a SMART

SMART Data Analysis at point in time for continuous and binary outcomes

  • Analysis for Main effects

  • Analysis for AI estimation and comparison: weighted and replicated regression

  • Example from Literature

SMART Longitudinal Data Analysis for continuous and binary outcomes

  • Analysis for AI estimation and comparison: weighted and replicated regression

  • Example from Literature Cluster

SMART Designs & Analysis Sample Size Calculations

  • Main effects & AI comparison and estimation

Further Tailoring Adaptive Interventions

  • Introduction to Q-learning

Introduction to small sample SMART design and analysis

Grant Writing Tips & Resources

Registration Options

SMART Design and Analysis

  • Professional
  • $499
  • Baseline Price for Faculty,
    Staff, and Other Professionals
  • Click Register Below
  • Trainee
  • $499 $334
  • 33% Discount for
    Students and Postdocs
  • Use code "TRAINEE" at Checkout

Combo 1: SMART Design and Analysis + MOST Framework

  • Professional
  • $748 $598
  • Baseline Price for Faculty,
    Staff, and Other Professionals
  • 20% Combination Discount
  • Click Register Below
  • Trainee
  • $598 $401
  • 33% Discount for
    Students and Postdocs
  • 20% Combination Discount
  • Use code "TRAINEE" at Checkout

Combo 2: SMART Design and Analysis + Pragmatic and Cluster Trials

  • Professional
  • $748 $598
  • Baseline Price for Faculty,
    Staff, and Other Professionals
  • 20% Combination Discount
  • Click Register Below
  • Trainee
  • $598 $401
  • 33% Discount for
    Students and Postdocs
  • 20% Combination Discount
  • Use code "TRAINEE" at Checkout

Note: All registration options for this workshop come with three things:
(1) Access to the video recording and materials of the 2024 version of the workshop until June 16, 2024
(2) The ability to attend the live recording of the 2025 version of the workshop on June 16-17, 2025
(3) Access to the video recording and materials of the 2025 version of the workshop after June 17, 2025

If this workshop is offered again in future years (e.g., 2026+), then you will have continued “evergreen” access to the new recordings and materials.

 FAQs

  • This workshop was custom built for researchers in the social, behavioral, medical, educational, implementation, and/or statistical sciences who are interested in SMART clinical trials. It is designed to be accessible to such learners with varying degrees of statistical and coding backgrounds and skills (from none to expert). This workshop is introductory and appropriate for faculty, post-docs, staff, and graduate students.

  • All levels

  • No prior experience is necessary, but some experience with clinical studies is encouraged.

  • We will illustrate the methods using R software. SAS code is available upon request. Applets will be provided for sample size calculations.

    If you aren't interested in coding, this workshop is still for you as we focus more on matching analyses to aims and present example results from applying methods to funded, published trials.

  • Slides, code, simulated dataset, reading list